Nonlinear total variation based noise removal algorithms

By L. Rudin, Stanley J. Osher, E. Fatemi. 1992

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

GAN Inversion: A Survey (2021, methodology): “It has been widely adopted in computer vision tasks, including MRF [196], [197], [198], dark channel prior [199] and total variation regularizer [200], which all model correlation among neighboring pixels via Gibbs distribution.”
High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs (2017, methodology): “We first obtain our results with adversarial training only, without relying on any hand-crafted losses [43] or pre-trained networks (e.”
Unsupervised Cross-Domain Image Generation (2016, background): “The last loss, L TV is an anisotropic total variation loss (Rudin et al., 1992; Mahendran & Vedaldi, 2015), which is added in order to slightly smooth the resulting image.”
Feature Denoising for Improving Adversarial Robustness (2018, methodology): “[8] transform the images via non-differentiable image preprocessing, like image quilting [4], total variance minimization [17], and quantization.”
Spatially Transformed Adversarial Examples (2018, methodology): “…arbitrary pixel p and its neighbors q ∈ N ( p ) , we enforce the locally smooth spatial transformation perturbation L ﬂow based on the total variation (Rudin et al., 1992): Intuitively, minimizing the spatial transformation can help ensure the high perceptual quality for stAdv, since adjacent…”
Countering Adversarial Images using Input Transformations (2018, methodology): “An alternative way of removing adversarial perturbations is via a compressed sensing approach that combines pixel dropout with total variation minimization (Rudin et al., 1992).”
Interval Coded Scoring: a toolbox for interpretable scoring systems (2018, background): “This penalty term corresponds to Total Variation Minimization (Rudin et al., 1992).”
Additive models with trend filtering (2017, background): “2005), which is known as 1d total variation denoising in signal processing (Rudin et al. 1992).”
Discovering governing equations from data by sparse identification of nonlinear dynamical systems (2015, methodology): “To counteract differentiation error, we use the total variation regularization (32) to denoise the derivative (33).”
Multi-View Silhouette and Depth Decomposition for High Resolution 3D Object Representation (2018, positive): “i,j √ (xi+1,j − xi,j)(2) + (xi,j+1 − xi,j)(2) [35] within the predicted ODM.”
EnhanceNet: Single Image Super-Resolution Through Automated Texture Synthesis (2016, background): “However, convolution transpose layers have been re-ported to produce checkerboard artifacts in the output, ne-1 cessitating an additional regularization term in the output such as total variation [43].”
The Dantzig selector: Statistical estimation when P is much larger than n (2005, neutral): “, where D 1 is the ﬁnite diﬀerence D 1β˜= β(t 1,t 2) −β(t 1 −1,t 2) and D 2β˜= β˜(t 1,t 2) − β˜(t 1,t 2 − 1); in short, kβ˜k BV is the ℓ 1-norm of the size of the gradient Dβ˜=(D 1β,D˜ 2β˜); see also [34]. Our example follows the data acquisition patterns of many real imaging devices which can colle”
Optimization with Sparsity-Inducing Penalties (2011, background): “Originally introduced in the image processing community [118], the total-variation penalty encourages piecewise constant signals.”
Learning Affinity via Spatial Propagation Networks (2017, methodology): “Other methods, such as total variation (TV) [23] and learning to diffuse [18] improve the modeling of pairwise relationships by utilizing different objectives, or incorporating more priors into diffusion partial differential equations (PDEs).”
Image inpainting (2000, background): “Finally, let us note that the choice of a non-normaliz ed field !N instead of a normalized version of it allows for a simpler and more stable numerical scheme; see [17, 18].”
Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising (2016, background): “Over the past few decades, various models have been exploited for modeling image priors, including nonlocal self-similarity (NSS) models [1], [2], [3], [4], sparse models [4], [5], [6], gradient models [7], [8], [9] and Markov random ﬁeld (MRF) models [10], [11], [12].”